[R] ldcase residuals

[Gareth Hughes]

Another Surv question:
According to the reference given in R help(Escobar and Meeker), 
'ldcase' approximates the change in minus twice the maximised 
log-likelihood on omission of each observation in turn. The 
maximum such residual is 1.26, but the actual likelihood 
displacement on omitting the correspondong observation is 7.37. 
Am I missing something?
Thanks